Using Wavelet-based Fractal Analysis of Inertial Measurement Unit Signals to Examine Gait Data from Men and Women during a Load Carriage Task

In the military, gait cycle rhythms during loaded marching have been shown to be neither random nor regular but instead exhibit a fractal temporal composition. Through wavelet-based fractal analysis, this study aimed to identify the complex fractal behavior of the gait stride interval dynamics of men and women during loaded ruck marching. We used data on the acceleration and angular velocity signals in three orthogonal directions (medial-lateral [ML], vertical [VT] and anteriorposterior [AP]) acquired from inertial measurement unit (IMU) sensors placed on six lower limb locations. To evaluate the stride-to-stride fractal patterns, we used wavelet transform and calculated the power spectral density (PSD). Thirty-three healthy adults (17 men and 16 women; age $=26.7\pm 5.9$ and $25.2\pm 4.5$ years, respectively) participated in this study and completed a 2-km best-effort loaded ruck (run or walk) marching task under field conditions. The results demonstrated that the acceleration signals from both the men and women are either anti-correlated or white noise $(0\leq\beta\leq 0.50)$, and this finding was observed across all directions. Most of the results from angular velocity signal was also found to be anti-correlated or white noise, with the exception that a few cases presented longrange correlation $(0.59\leq\beta\leq 1.0)$; for example, the angular velocity signal in the AP direction obtained from all six IMU placements showed a long-range correlation for both sexes. The fractal dimensions showed significant differences $(\mathrm{p}\lt 0.05)$ between the directions (i.e., ML vs. VT, ML vs. AP, and AP vs. VT) of both signals for both sexes. These results might be beneficial for detecting gait abnormalities in military men and women, which would help identify individuals at greater risk for injury and aid the evaluation of interventions, specifically designed to minimize said injury risk(s).